Why do you have to use both a compiler flag and a run-time flag to get multicore-support in Haskell?

views:

238

answers:

+12 Q:

Why do you have to use both a compiler flag and a run-time flag to get multicore-support in Haskell?

+3 A:

From GHC User guide (version 6.12.1):

Omitting x, i.e. +RTS -N -RTS, lets the runtime choose the value of x itself based on how many processors are in your machine.

I suppose there's no specific reason for this not to be the default, apart from authors' vision of what should defaults be. (Note that this also enables parallel GC, which maybe sometimes isn't what you wish to be by default.)

Roman Cheplyaka 2010-09-13 21:48:03

Thanks for pointing that out. It addresses the "number of cores" portion of my question. That is a start. However, I'm more frustrated by having to set flags both to compile AND to run my code.

Tim Perry 2010-09-13 22:10:00

While you're developing the program this shouldn't be a big deal and for the final (shipped) binary you can link it with some [static RTS options](http://haskell.org/ghc/docs/6.12.2/html/users_guide/runtime-control.html#rts-hooks) via a .h file with `char *ghc_rts_options = "-N";`

TomMD 2010-09-13 22:44:37

@TomMD: That is a useful hack. It would solve the issue I'm currently fighting. If you post that as a solution, I'll accept it. I'm also planning to submit a feature request as Don Steward suggested above.

Tim Perry 2010-09-13 22:58:16

+6 A:

Why can't the run-time executable detect it was compiled with -threaded and use all cores on the system unless otherwise specified?

That's an interesting feature request!

You could ask for it on the GHC feature tracker: http://hackage.haskell.org/trac/ghc/wiki/ReportABug

Don Stewart 2010-09-13 22:21:17

Will submit. Thanks for hint.

Tim Perry 2010-09-13 22:58:41

Submitted. Ticket #4307.

Tim Perry 2010-09-13 23:06:57

Considering that you can do what TomMD suggested should this bug request be considered closed?

Robert Massaioli 2010-09-14 05:12:43

I don't think so. After all, I could hack the entire program together in assembly if I wanted to. But I don't want to. I want compiling with -threaded to provide behavior currently enabled by placing "+RTS -N -RTS" on the command line. I think it will be much cleaner that way.

Tim Perry 2010-09-14 18:02:10

I agree with Tim on this. While no programmer should be too put-off by having an extra header file, a little more polish doesn't hurt here.

TomMD 2010-09-14 23:01:42

So I submitted it and it was closed. Apparently making the runtime less obscure is not a priority of the compiler team. Re-submitted to the runtime team. Hoping for different result. http://hackage.haskell.org/trac/ghc/ticket/4319

Tim Perry 2010-09-15 21:32:53

+10 A:

While you're developing the program the extra +RTS ... shouldn't be a big deal (though I admit it struck me as odd when I first picked up Haskell). For the final (shipped) binary you can link it with static RTS options (GHC manual) by providing a C file containing char *ghc_rts_opts = "-N";.

TomMD 2010-09-13 23:50:29

At least, there is a hook. But what an ugly way to specify default RTS options.

gawi 2010-10-04 00:24:26

ansaurus

tags:

views:

answers:

Why do you have to use both a compiler flag and a run-time flag to get multicore-support in Haskell?

related questions